A workable ad creative refresh schedule replaces half of each live set on a fixed cadence: weekly for TikTok and app install, fortnightly for Meta feed and Reels, monthly for long-cycle B2B. Set the cadence by what you can actually produce, not by what the decay curve deserves, because a schedule you cannot supply collapses in week three and leaves you running the same tired ad you were running before you made the plan. Cost it in production hours and credits first, then commit.
Which cadence does each placement need
Decay speed is a property of the placement and the vertical, not of your ambition. Our 2026 fatigue benchmark puts days to a 40 percent CTR decline at 9 days for food and beverage, 12 to 14 for fashion, about 18 for beauty and DTC, about 21 for electronics and about 28 for B2B SaaS. Placement ceilings differ too: 2.5 weekly frequency on Meta feed and Reels, about 3.0 on TikTok, 3 to 7 per week on YouTube in-stream.
Put those together and the cadence falls out. Weekly on TikTok and app install, where both the decay and the frequency accumulation are fastest. Fortnightly on Meta feed and Reels. Monthly is defensible for B2B SaaS on LinkedIn, where the same benchmark gives you about 28 days before a 40 percent decline, and where the audience is small enough that a different problem dominates.
A food and beverage brand on TikTok and a B2B SaaS on LinkedIn should not run the same schedule, and the gap between 9 days and 28 days is the reason.
What each cadence costs to supply
This is the table most refresh-schedule advice leaves out, and it is the only one that determines whether the plan survives.
| Cadence | Fits | Live set | Replaced per cycle | New variations per month | Credits per month | Brief and review time |
|---|---|---|---|---|---|---|
| Weekly | TikTok, app install, food and beverage | 12 | 6 | About 26 | 26 | 6 to 7 hours |
| Fortnightly | Meta feed and Reels, fashion, DTC | 12 | 6 | 13 | 13 | 3 to 4 hours |
| Every three weeks | Electronics, considered purchases | 10 | 5 | About 7 | 7 | 2 hours |
| Monthly | B2B SaaS, LinkedIn, small audiences | 8 | 4 | 4 | 4 | 1 hour |
| Quarterly | Nothing, and this is the point | 8 | 8 | About 3 | 3 | Under an hour |
The credits column assumes one standard video variation costs 1 credit, which is what it costs on our plans. The hours column is our experience of brief-plus-review once a footage library is already uploaded and tagged, at roughly 15 minutes of human attention per variation. It is not shoot time, because in this model there is no shoot.
Read the last row as a warning rather than an option. Brands shipping 15 to 50 creative variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers, according to our benchmark report, and the report's blunt operational conclusion is that throughput rather than talent is the bottleneck.
Then compare the credit column against our pricing. The fortnightly row at 13 variations a month sits inside a 15-credit pack. The weekly row at 26 does not, and the 65-credit Growth pack at 99 euros covers about two and a half months of it. Deciding that before you publish the calendar is what stops the calendar being fiction. Our ad variation calculator will size the set for your own placement mix if you want the number rather than the table.
Why you replace half the set, not all of it
Rotating an entire set at once is the most common way a good cadence produces bad results.
Three things go wrong. You throw away delivery history, so the algorithm has nothing to concentrate spend on and every ad in the set starts cold at the same time. You lose the comparison, because a new set measured against another new set tells you nothing about whether the change helped. And you create a sawtooth: performance dips for two or three days after every rotation, which on a weekly cadence means you spend a third of every month in recovery.
Replace the bottom half instead. Rank the live set by hook rate against each ad's own first week, retire the bottom rows, and let the top rows keep their learning until they cross the frequency ceiling. The set stays warm, spend keeps flowing to something proven, and you still get a full cycle of new concepts into the auction. This is the rhythm the creative testing workflow we run is built around, and it is more forgiving of a missed week than a full rotation is.
One exception. If the whole set shares a claim that has stopped working, replacing half of it just re-runs the same failure. That is a concept problem, not a fatigue problem, and it earns a full reset.
The warm bench
The single change that makes a refresh schedule survive contact with a bad week: never have only one winner.
Keep a runner-up live at 10 to 15 percent of the ad set budget. It is not there to win. It is there to have delivery history, a stable hook rate and an audience it has not exhausted, so that on the Tuesday your best performer falls off the cliff you promote a known quantity in ten minutes instead of briefing a replacement and waiting three days for learning. The cost of a warm bench is a small tax on efficiency. The cost of not having one is the gap between a winner dying and a replacement being ready, which for most teams is a week of paying full price for declining creative.
Keep two or three bench candidates rather than one, as a subset of the 8 to 20 live variations an active campaign should carry anyway.
What the month actually looks like
A fortnightly Meta schedule for a DTC account with 12 live variations, run as a repeating four-week block:
- Week 1, Monday: rank the set, retire the bottom 6, promote the two strongest bench candidates into the main rotation, and brief 6 replacements. Six credits.
- Week 1, Thursday: review and publish the 6 new variations at low budget. Two of them are earmarked as next cycle's bench.
- Week 2: no production. Read hook rate against each ad's own first week and let delivery concentrate.
- Week 3, Monday: same as week 1. Rank, retire 6, brief 6. Six more credits.
- Week 3, Thursday: publish.
- Week 4: no production. Check frequency at ad level, not campaign level, and mark anything over 2.5 for retirement in the next cycle.
That is 12 credits and about three and a half hours of human attention a month, on a rhythm you can run in a week where something else has caught fire. The weekly version is the same block twice as often, at roughly 26 credits a month.
Where we fit, plainly
Genyad exists to make the supply column above achievable without a shoot. You upload footage once, it transcribes and tags every clip, and each variation is a new script, shot selection, voiceover, caption set and export drawn from that library rather than a re-cut of one timeline. Exports come in 9:16, 4:5, 1:1 and 16:9, and editing or re-exporting an existing variation costs nothing. Credits never expire, which matters on a cadence where a quiet month should not burn a subscription.
What it does not do: there are no AI avatars or synthetic presenters, no static banner formats, no product-URL import, no product-feed or CSV template rendering, no predicted performance scores, and no direct publishing to Meta or TikTok. Every schedule above assumes a human uploads the exports and reads the results in the platform. If you have no usable footage at all, an avatar or generative tool is a better fit than we are.
Frequently asked questions
How often should ad creative be refreshed?
Weekly for TikTok and app install, fortnightly for Meta feed and Reels, monthly for long-cycle B2B. Our benchmark report puts days to a 40 percent CTR decline at 9 for food and beverage, 12 to 14 for fashion, about 18 for beauty and DTC, about 21 for electronics and about 28 for B2B SaaS, so pull your cadence toward the faster end if your vertical sits there.
How many new variations does a refresh schedule need per month?
Between 4 and 26 depending on cadence and set size, replacing half of an 8 to 20 variation rotation each cycle. Brands shipping 15 to 50 variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers in our benchmark report.
Should I replace all of my ads at once?
No, unless the whole set shares a claim that has stopped working. Replacing everything resets delivery history for every ad simultaneously, which produces a performance dip for a few days after each rotation and removes any basis for comparison. Retire the bottom half by hook rate and let proven ads keep running until they cross the frequency ceiling.
What if I cannot keep up with a weekly cadence?
Run fortnightly with a warm bench rather than weekly with gaps. A cadence you hit every time beats a faster one you miss, because the failure mode of a missed cycle is running exhausted creative at full budget. Pick the fastest cadence you can actually supply.